Staging Systems are Overrated - Docker-Based System Testing
summary
Claims concrete results: a repo-level JSON describes providers and from that generates valid docker setup scenarios and per-repo Docker images, Jenkins builds push versioned images every commit and a job re-computes dependency graphs every five minutes. System runs on Mesos/Marathon with Kafka and Storm (exactly-once semantics), uses spot and prepaid tiers, handles 400M images and ~50B objects, and turns integration tests into system tests materialize all downstream dependencies so CI exercises real Postgres/Redis/Elasticsearch/Kafka stacks and actual deployable binaries. Boot order and config are enforced, not handwaved.
AI-written summary · about
chapters (11)
- 0:00Start
- 2:46Semantic Segmentation
- 3:183d Reconstruction
- 6:28Docker Base Deployment
- 7:28Stateless Scaling
- 8:08Apache Storm
- 8:40Kafka Topic
- 15:07Monitoring
- 15:30How Does Testing Work
- 30:58Docker Providers
- 45:43Python
related talks
38mWhy You Need To Stop Using "The" Staging Server
36mWhy You Need to Stop Using 'THE' Staging Server
44mDocker and the Path to a Better Staging Environment
52mContinuous Delivery with Containers: The Good, the Bad, and the Ugly
32mUsing Docker in test automation
33mDocker Container Lifecycles – Problem or Opportunity?
49mContinuous Delivery at Docker age
43mLightweight PaaS for Jenkins CI Environments with Docker